Plasma glutamine concentration after intensive care unit discharge: an observational study

A low plasma concentration of glutamine at intensive care unit (ICU) admission is associated with an unfavorable outcome [1],[2]. A high glutamine concentration at ICU admission may also communicate a negative prediction [2]. Other reports claim that free glutamine depletion in tissues (during ICU stays) is associated with unfavorable outcome [3],[4]. So the hypothesis - that at least some critically ill patients may have a shortage of free glutamine - seems to be valid. Conventional products for enteral feeding of critically ill patients contain the same amounts of glutamine as ordinary food, while conventional products for parenteral feeding of these patients do not contain any glutamine, related to stability issues. Consequently, supplementation of glutamine was suggested [5].

Literature on glutamine supplementation for critically ill patients is extensive. Meta-analyses suggest a beneficial effect when intravenous (iv) supplementation is given to ICU patients on parenteral nutrition [6],[7]. Reports on enteral supplementation to ICU patients on enteral nutrition are less conclusive [6]. Only a few studies include plasma glutamine concentration measurements as a possible indicator of individuals with glutamine depletion, which should be the proper target group for supplementation [8]-[10]. The rational for supplementing critically ill patients - without knowledge of their actual glutamine status - has been that glutamine supplementation was considered to be without side effects [5],[11]. But recently, a study demonstrated harm when supplying glutamine in high doses as a part of a pharmaconutrition concept in the acute phase of critical illness [9], although the subjects were a selected subgroup that received hypocaloric nutrition and had two or more organ failures. So better knowledge of the glutamine status and whether or not an actual deficiency of glutamine is present in the individual patient is needed.

The Scandinavian Glutamine Trial indicated in secondary outcomes of only patients treated per protocol that iv glutamine supplementation may improve outcome during ICU stay [12]. This possible benefit, however, was not sustained after ICU discharge, which may be related to that glutamine supplementation was discontinued at discharge. To elucidate this hypothesis two observational research studies were launched: Study A - to follow plasma glutamine concentrations after ICU discharge, and Study B - to find out if post-ICU plasma glutamine concentration is a mortality predictor.

In study A (the post-ICU temporal pattern of plasma glutamine concentrations) patients were included between October 19, 2011 and June 11, 2012 (Figure 1A). Sixty-three patients were included to reach the target, which resulted in 49 patients who could be evaluated for their post-ICU plasma glutamine concentrations. For patient characteristics see Table 1, and for ICU discharge, total amino acids and glutamine concentrations see Table 2. Figure 2 shows the plasma glutamine concentration in each patient. In subpanels B and C patients are divided into (i) those who were readmitted or not readmitted to ICU, and (ii) survivors and nonsurvivors. The glutamine supplementation periods are indicated for readmitted patients. In total, 210 post-ICU samples were collected: 6 were <400 μmol/L and 16 were >930 μmol/L. Values outside these limits have been shown to be associated with an increased mortality risk at ICU admission [2]. At discharge from the ICU, plasma glutamine concentration dropped in 30 out of 49 patients from 690 (532,818) to 622 (506,765) μmol/L (P = 0.054).

Regarding the samples collected during ongoing iv glutamine supplementation on the last day of ICU stay all 63 patients were evaluated (Tables 1 and 3a,b), there was a difference between 12-month survivors and nonsurvivors, 596 (491,744) versus 777 (648,848) μmol/L (P = 0.004) respectively (Figure 3A). There was also a difference in the sequential organ failure assessment (SOFA score on the last day of ICU stay 3 (2,5) versus 7 (4,9) (P = 0.0003). Using univariate logistic regression analysis, discharge SOFA score, admission simplified acute physiology score III (SAPS III) score, dialysis, and discharge plasma glutamine concentration were identified as mortality predictors (Table 3a). The stepwise regression analysis indicated that at discharge, the discharge SOFA score complemented the discharge plasma glutamine concentration and enabled the best mortality prediction (Table 3b).

In study B (the post-ICU concentration of plasma glutamine as outcome predictor), patients were included between October 10, 2011 and March 27, 2013 (Figure 1B). The target was 100 patients, out of which 92 could be evaluated, for patient characteristics see Table 1, and for post-ICU total amino acids and glutamine concentrations see Table 2. No difference in post-ICU plasma glutamine concentration existed regarding 12-month, all-cause mortality: 490 (397,604) versus 517 (418,675) μmol/L (P = 0.45), for survivors (n = 54) and nonsurvivors (n = 38) respectively (Figure 3B). Post-ICU plasma glutamine concentration was not a significant mortality predictor (Table 3c). In addition, there was no significant correlation between post-ICU plasma glutamine concentration and discharge SOFA score, r = 0.07, (P = 0.51). Using univariate regression analysis, discharge SOFA score as well as admission SAPS III were identified as mortality predictors, but not post-ICU plasma glutamine concentration (Table 1c). The stepwise regression analysis indicated that neither discharge SOFA score nor post-ICU plasma glutamine concentration added any prediction value to admission SAPS III score, which gave the best mortality prediction in this group of patients.

The main findings from studies A and B were that post-ICU plasma glutamine levels were within the normal range and that they did not predict mortality. Not unexpectedly, admission SAPS III and discharge SOFA were the strongest mortality predictors in this highly selected subgroup of ICU patients. Remarkably, discharge plasma glutamine concentration during ongoing intravenous supplementation added prediction value - with a higher concentration within normal limits being more unfavorable.

The nutrition protocol was adapted from the protocol used in the Scandinavian Glutamine Trial [12] because hypotheses behind the present protocols were generated from that trial. This, of course, limits the generalizability of the findings. Eligible patients were supplemented intravenously with a glutamine-containing dipeptide. The objective was to normalize plasma glutamine levels. Study A somewhat demonstrated that this objective was achieved. Out of 63 patients sampled on the last day of ICU stay during ongoing glutamine-containing infusing, 56 out of 63 were within the range 400 to 930 μmol/L, 3 were <400 μmol/L and 4 were >930 μmol/L. These limits were chosen from an earlier study when glutamine levels at admission were tested for outcomes prediction [2]. Sometimes the normal range in healthy subjects is given as 500 to 850 μmol/L [13], but we thought the reference to limits that were proven to represent risk at admission (among the studied mixed cases) may be of greater relevance.

The hypothesis of post-ICU plasma glutamine concentration as a mortality predictor could be dismissed in the patient group that required a long ICU stay. Admission SAPS III score was the best mortality predictor in this selected group of patients, also at discharge. From these data, there is no rational to treat patients with glutamine supplementation post ICU. In general, post-ICU plasma glutamine levels were not low, and they did not facilitate mortality prediction. Yet patients with a glutamine deficiency might exist, and determination of plasma glutamine concentrations (or even better, determination of endogenous glutamine production) should identify such patients. Studies in this area might resolve this issue.

Supraphysiological plasma glutamine concentrations are perhaps more common in critically ill patients than earlier thought. In our observational study of admission glutamine levels, concentrations >930 μmol/L were associated with an unfavorable outcome [2]. In acute fulminant liver failure similar supraphysiological glutamine levels are quite common [14]. In an earlier publication, we reported a case with an extremely high concentration, which was normalized upon recovery [15].

In studies A and B, 16 out of 210 post-ICU samples were supraphysiological. Out of these, 11 were during iv glutamine supplementation in the ICU after readmission, and 5 were post ICU without any supplementation. Of the supraphysiological values 7 out of 16 values were recorded in 3 patients who survived and, 9 out of 16 values were recorded in 4 patients who did not survive (Figure 2C). Most importantly, studies are needed to explain the relation between plasma glutamine concentration and the endogenous glutamine production in various states of critical illness. Others have reported a higher plasma glutamine concentration (within normal limits) in nonsurviving long-staying ICU patients, but supraphysiological concentrations of other amino acids were reported to be stronger mortality predictors in that report [16].

Studies A and B were not designed to find out if supraphysiological levels of glutamine may be toxic. The REDOX study reported a more unfavorable outcome in ICU patients - given a hypocaloric diet containing pharmacological doses of glutamine administered enterally and intravenously [9]. But the plasma concentration data from a small subgroup of these patients did not reveal high glutamine values during treatment. So the harm associated with glutamine supplementation in the REDOX study was not associated with high plasma glutamine concentrations in the reported data. In the material from studies A and B, there is a signal from individual subjects that high glutamine levels - regardless of supplementation - may be associated with an unfavorable outcome. More clarity is needed whether plasma glutamine concentration is an indicator of glutamine depletion. Consequently, further studies over glutamine kinetics (production and utilization) in relation to plasma glutamine concentration are needed [17].

Particular interest has been given to kidney failure as a possible association to high glutamine concentrations and an unfavorable outcome [7]. Here we found dialysis to be a mortality predictor at ICU discharge in Study A. However, this statistical relationship was not maintained in the multivariate analysis outside the prediction contained in the discharge SOFA score. It must be emphasized that these post hoc analyses are of a purely hypothesis-generating character, and the statistical power contained in these calculations is limited [18],[19]. In particular, it may be noted in Table 1 the difference in dialysis at discharge as related to mortality between groups A and B, although there were identical inclusion criteria and a considerable overlap in recruitment between the two groups, underlining the hazards associated with post hoc analyses.

The strengths of studies A and B are that consecutive patients were included and that 100% follow-up was possible. The limitations are that the ICU patient case mix may limit the generalizability of the result. Ongoing intravenous glutamine supplementation during ICU stays is also uncommon, but was part of the hypothesis following the result of the Scandinavian Glutamine Trial [12].

ICU survivors were not low in plasma glutamine after ICU discharge, although no exogenous supplementation was given. In addition, post-ICU glutamine level was not a mortality predictor in the way that admission glutamine has been demonstrated to be. So after ICU discharge, exogenous glutamine supplementation does not have the same rationale as during ICU stay. At ICU discharge a higher plasma glutamine level was associated with post-ICU mortality. The possible indication for exogenous glutamine supplementation during ICU stay is still not properly defined.

Marie Smedberg, MD PhD student at the Division of Anaesthesia at CLINTEC, Karolinska Institutet, and resident at the Department of Anesthesia and Intensive Care Medicine, Karolinska University Hospital Huddinge.

Johanna Nordmark Grass, MD PhD, Postdoctoral Fellow at the Division of Anaesthesia at CLINTEC, Karolinska Institutet, and senior consultant at the Department of Anesthesia and Intensive Care Medicine, Karolinska University Hospital Huddinge.

Linn Pettersson, MD, resident at the Department of Anesthesia and Intensive Care Medicine, Karolinska University Hospital Huddinge.

Åke Norberg, MD PhD, Postdoctoral Fellow at the Division of Anaesthesia at CLINTEC, Karolinska Institutet, and senior consultant at the Department of Anesthesia and Intensive Care Medicine, Karolinska University Hospital Huddinge.

Olav Rooyackers, PhD, Professor of Anaesthesia and Intensive Care Medicine at the Division of Anaesthesiology at CLINTEC, Karolinska Institutet.

Jan Wernerman, MD PhD, Professor of Anaesthesiology and Intensive Care Medicine at the Division of Anaesthesia at CLINTEC, Karolinska Institutet, and senior consultant at the Department of Anesthesia and Intensive Care Medicine, Karolinska University Hospital Huddinge.